We have developed a mechanistic model to interpret spectral induced polarization data of partially saturated clay-rocks. This model accounts for the polarization of the grains through an electrical double layer model with a polarization model of the inner part of the electrical double layer called the Stern layer. The polarization model accounts also for the Maxwell–Wagner polarization at frequencies higher than 100 Hz. The Maxwell–Wagner polarization is modelled by using a conductivity model modified to account for the presence of a non-wetting immiscible phase like air in the pore space. The resulting model is consistent with the first and second Archie's laws in the case where surface conductivity can be neglected. The volumetric charge density of the diffuse layer at saturation is divided by the saturation of the water phase to account for the partial water saturation of the porous material. The model comprises seven fundamental parameters: the formation factor, the second Archie's exponent, a critical water saturation level, the mean electrical potential of the pore space at saturation, the density of the counterions in the Stern layer, and at least two parameters describing the grain size distribution. Most of these parameters can be derived independently using alternative measurements and electrochemical models. Measurements were performed in the frequency range 10 mHz–45 kHz using five samples from the Callovo–Oxfordian formation in the eastern part of the Paris Basin, France. The model agrees fairly well with the experimental data at saturation and for partially saturated clay-rocks down to 1 Hz. Most of the seven physical parameters entering the model were independently evaluated.